Abstract: The key step to construct the system of automatic iris recognition is how to locate the iris quickly. To achieve such goal, according to the characteristics of the iris image, an algorithm, which is based on block statistics of average grey, is proposed to determine the centre and radius of the iris inner circle. Using moment invariants on edge regions, the parameters symbols of the iris outer circle can be recognized conveniently and quickly, avoiding the use of accumulated count of any parameters. Algorithms proposed in this paper have been simulated on 50 sets of images from CASIA iris image database and the locating accurate rate is up to 95%.

Key words: iris location, gray mean, block, moment invariants

摘要： 快速虹膜定位是实现虹膜自动识别系统的基础。从虹膜图像特点出发，提出了基于分块统计灰度平均值的方法来确定虹膜内圆圆心和半径。利用不变矩的识别方法，不需要对任何带估参数进行累积计数，就可快速求出外边缘的参数，计算量小，计算时间短。通过对中科院自动化所CASIA虹膜数据库50组图像的虹膜定位测试结果表明，该方法定位准确率达到95%。

关键词: 虹膜定位, 灰度平均, 块, 不变距